Suppose，again，thathehasacquiredmoreexperience，andhaslivedsolongintheworldastohaveobservedfamiliarobjectsoreventstobeconstantlyconjoinedtogether；whatistheconsequenceofthisexperience？Heimmediatelyinferstheexistenceofoneobjectfromtheappearanceoftheother。Yethehasnot，byallhisexperience，acquiredanyideaorknowledgeofthesecretpowerbywhichtheoneobjectproducestheother；norisitbyanyprocessofreasoning，heisengagedtodrawthisinference。Butstillhefindshimselfdeterminedtodrawit:
Andthoughheshouldbeconvincedthathisunderstandinghasnopartintheoperation，hewouldneverthelesscontinueinthesamecourseofthinking。Thereissomeotherprinciplewhichdetermineshimtoformsuchaconclusion。
ThisprincipleisC/USTOM\\orH/ABIT\\。Forwherevertherepetitionofanyparticularactoroperationproducesapropensitytorenewthesameactoroperation，withoutbeingimpelledbyanyreasoningorprocessoftheunderstanding，wealwayssay，thatthispropensityistheeffectof。Byemployingthatword，wepretendnottohavegiventheultimatereasonofsuchapropensity。Weonlypointoutaprincipleofhumannature，whichisuniversallyacknowledged，andwhichiswellknownbyitseffects。
Perhapswecanpushourenquiriesnofarther，orpretendtogivethecauseofthiscause；butmustrestcontentedwithitastheultimateprinciple，whichwecanassign，ofallourconclusionsfromexperience。Itissufficientsatisfaction，thatwecangosofar，withoutrepiningatthenarrownessofourfacultiesbecausetheywillcarryusnofarther。Anditiscertainwehereadvanceaveryintelligiblepropositionatleast，ifnotatrueone，whenweassertthat，aftertheconstantconjunctionoftwoobjects——heatandflame，forinstance，weightandsolidity——wearedeterminedbycustomalonetoexpecttheonefromtheappearanceoftheother。Thishypothesisseemseventheonlyonewhichexplainsthedifficulty，whywedraw，fromathousandinstances，aninferencewhichwearenotabletodrawfromoneinstance，thatis，innorespect，differentfromthem。Reasonisincapableofanysuchvariation。Theconclusionswhichitdrawsfromconsideringonecirclearethesamewhichitwouldformuponsurveyingallthecirclesintheuniverse。Butnoman，havingseenonlyonebodymoveafterbeingimpelledbyanother，couldinferthateveryotherbodywillmoveafteralikeimpulse。
Allinferencesfromexperience，therefore，areeffectsofcustom，notofreasoning。[12]